6-29-22 Sakuradai, Nerima-ku, Tokyo 176-0002, Japan
The style of speech illustrated is that of many educated Japanese brought up in Tokyo or
other areas with similar pitch accent systems. The transcription is based on a recording of a
25-year-old student whose speech is typical of speakers of his age group with this
gaito: 'cloak, overcoat
(See also /b, p, d, r7.)
zat s uzi
i imi 'meaning'
e 6me 'smile!' (archaic imperative)
a ama 'woman diver (for abalone)'
o 6mo '(sur)face' (archaic)
u umu 'suppurate'
1 1 8 Handbook of the IP A
ozi: santafi ' grandfathers '
hasi o nuru 'paint chopsticks'
hasi o nuru 'paint a bridge'
hasi o nuru 'paint the end'
/p, t, t s , k/ are only moderately aspirated. Syllable-final (moraic) forms of these and other
voiceless obstruents occur as the first part of geminates, e.g. /happo:/ 'firing', /jatto/ 'at
last', /gakko:/ 'school', /hossa/ 'attack (of disease)'. The geminate form of /t7 is /tt7, as in
/kutfukemasita/ 'pressed, attached', /b/ is normally [b], but in rapid speech it may become
[v] or [£]. Igl tends to become [y] between vowels. Older and otherwise conservative
speakers use [rj] in all medial positions, except mainly for the element /g67 'five' in number
designations, as e.g. /nizju:-go/ [jjizuigo] 'twenty-five'. /rV, which corresponds to 'r' in
Romanization, is postalveolar in place rather than retroflex and mainly occurs medially.
Initially and after /n/, it is typically an affricate with short friction, [dl]. A postalveolar [I] is
not unusual in all positions. Approximant [j] may occasionally occur in some environments.
/t7 is normally [ts] but becomes [tc] before I'll, /n/ before HI is prepalatal }]. Before /u/, it
may be alveolar with some speakers. Consonants generally are strongly palatalized before
/i/, as /mi/ [mH] 'body', /n/ represents a moraic nasal with very variable pronunciation.
Word-finally before a pause, it is typically a uvular nasal with a loose tongue contact or a
close to close-mid nasalized vowel. Such a nasalized vowel is also the normal realization
before a vowel or approximant, or before /h/ or /s/, as e.g. Iz6ml [ d z£ii] 'goodwill',
/hoNJaku/ [hoijaku] 'translation', /zeNhaN/ [ d zefhaN] 'first half. Before other consonants,
it is homorganic with that following consonant, as e.g. /siNpo/ [cimpo] 'progress', /siNni/
[cijjjji] 'truly', /siNti/ [cind-H] 'truth', /s/ and /z/ are [c] and [z] before /i/. Ill tends to be
[ d z] initially ([ d z] before HI) and after /n/. Ihl tends towards  and [$] before HI and /u/
respectively. Ihhl is realized as , [xx] or [§§] depending on the (normally identical)
surrounding vowels. 1)1 affects the preceding consonant as HI does, and is itself absorbed,
thus: /mjakuV [mJaku] 'pulse', /t s ja/ [tea] 'tea', /sj6:/ [co:] 'prize' , /kanju:/ [kajju:] 'joining'.
/t,d/ before /i,j/, normally limited to borrowings, remain essentially unpalatalized, thus
/tiipaiti:/ [tiipaitii] 'tea party', /direkuta:/ [dir£kutai] 'director', /djiiaru/ [djuaru] 'dual',
/w/ has very slight or no rounding (except after I of), but involves no spreading either.
/u/, resembling [in] auditorily, has compressed lips, so that it is unrounded but without
spreading; it could be transcribed narrowly as [\f] or [W]. The slit between the lips may be
very narrow vertically and is generally much shorter in the horizontal plane than for [i]. A
very advanced variety in the [i] area may occur among the younger age groups. Except in
accented or lengthened syllables, /i, u/ tend to be devoiced [i, u] between voiceless
consonants. As often as not, preceding fricatives replace them altogether. Final unaccented
/si 1 / is very often reduced to [s].
Illustrations of the IP A 1 1 9
There are two lexically relevant pitch levels: high and low. Within a word, if the first
mora is high-pitched, the second is inevitably low-pitched, and vice versa. A mora
transcribed with an acute accent, a, is said to be accented and is high. If more than one mora
precedes it, the high pitch extends towards the beginning of the word up to the second
mora. A word with no accent mark begins low and continues high from the second mora
onwards. All moras following the accent are low until another accented or polysyllabic
word is reached, when these accentuation rules reapply. Note that word pairs such as /nasi/
'bridge' and /nasi/ 'end' are both low-high when spoken in isolation, but will be
distinguished when something else follows. What follows will be low after the accent, but
otherwise high. When a long (two-mora) vowel is transcribed with an accent, only the first
mora is high, and a pitch drop occurs between the two moras.
Transcription of recorded passage
arutoki kitakaze to taijo: ga t s ikarakutabe o simasita. tabibito no gaito: o
nugaseta ho: ga kat s l to ju: koto ni kimete, mazu, kitakaze kara
hazimemasita. kitakaze wa, nani, hitomakuti ni site misejo:, to, hagesiku
hukitatemasita. suruto tabibito wa, kitakaze ga hukeba hukuhodo gaito: o
sikkarko karada ni kutt s ukemasita. koNdo wa taijo: no baN ni narjmasita.
taijo: wa kumo no aida kara jasasii kao o dasite, atatakana hikari o
okurimasita. tabibito wa daNdaN joi kokoromot s i ni natte, simai ni wa
gaito: o nugimasita. sokode kitakaze no make ni narjmasita.
Romanization (Hepburn system)
Arutoki Kitakaze to Taiyo ga chikara-kurabe o shimashita. Tabibito no gaito o nugaseta ho
ga kachi to yu koto ni kimete, mazu Kitakaze kara hajimemashita. Kitakaze wa, 'Nani,
hitomakuri ni shite miseyo', to, hageshiku fukitatemashita. Suruto tabibito wa, Kitakaze ga
fukeba fukuhodo gaito o shikkarito karada ni kuttsukemashita. Kondo wa Taiyo no ban ni
narimashita. Taiyo wa kumo no aida kara yasashii kao o dashite, atatakana hikari o
okurimashita. Tabibito wa dandan yoi kokoromochi ni natte, shimai ni wa gaito o
nugimashita. Sokode Kitakaze no make ni narimashita.
Handbook of the International Phonetic Association :
A Guide to the Use of the International Phonetic Alphabet
International Phonetic Association, July 1999.